A data center cluster scheduling method for energy-saving control

Abstract

The present invention provides a data center cluster scheduling method for energy-saving control, which belongs to the field of data center technology, and specifically comprises: utilizing the historical processing conditions of similar processing tasks in various data centers to determine data center groups corresponding to different historical processing times, determining available groups in the data center groups based on the data interaction conditions between data centers in the historical processing times of the similar processing tasks in different data center groups, determining historical energy consumption data of different data centers in the available groups based on the composition data of the data centers in the available groups, and utilizing the historical energy consumption data to determine data processing groups in the available groups, performing data processing of data processing tasks through the data processing groups, and reducing the energy consumption of data processing.

Description

Technical Field

[0001] The present invention belongs to the technical field of data centers, and in particular relates to a data center cluster scheduling method for energy-saving control. Background Art

[0002] In order to achieve the scheduling of data center clusters and reduce the carbon emissions of data centers, the invention patent application CN202410336695.4 "A data center cluster scheduling method and device based on resource sharing" aims to minimize the energy cost of each data center, solves the resource scheduling model, and obtains the scheduling results of each data center, so that various resources of the data center cluster can be reasonably scheduled, energy saving and carbon reduction can be achieved, and resource waste can be avoided. However, the following technical problems are ignored:

[0003] In the process of scheduling and processing data centers, the existing technical solutions ignore the data exchange between different data centers. If the amount of data interaction between different data centers in the same data processing task is large, data errors or data transmission failures will inevitably exist in the data interaction process, making it difficult to meet the processing reliability requirements of the data processing task.

[0004] In response to the above technical problems, the present application specifically provides a data center cluster scheduling method for energy-saving control. Summary of the invention

[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0006] According to one aspect of the present invention, a data center cluster scheduling method for energy-saving control is provided.

[0007] A data center cluster scheduling method for energy-saving control, specifically comprising:

[0008] S1 determines the data volume of different data types corresponding to the data processing task based on the analysis result of the data processing task, and determines the similar processing tasks and task processing similarity coefficients of the data processing task in combination with the data processing requirements of different data types;

[0009] S2 determines the data processing delay and energy consumption of different similar processing tasks in each data center, and when it is determined that there is no data center that meets the data processing reliability requirements based on different task processing similarity coefficients, proceeds to the next step;

[0010] S3 determines data center groups corresponding to different historical processing times by using the historical processing conditions of the similar processing tasks in each data center, and determines available groups in the data center group based on the data interaction conditions between data centers in the historical processing times of the similar processing tasks in different data center groups;

[0011] S4 determines the historical energy consumption data of different data centers in the available group based on the composition data of the data centers in the available group, and uses the historical energy consumption data to determine the data processing group in the available group, and performs data processing of the data processing task through the data processing group.

[0012] The beneficial effects of the present invention are:

[0013] The available groups in the data center group are determined based on the data interaction situations between data centers in the historical processing times of the similar processing tasks in different data center groups, thereby avoiding the technical problem of insufficient data processing reliability due to the large number of corresponding data interaction times in the data center group and the large number of data centers with data interaction, reducing the frequency of data interaction between data centers and improving the reliability of data processing for data processing tasks.

[0014] The historical energy consumption data of different data centers in the available group are used to determine the data processing group in the available group, fully considering the energy consumption of different data centers under different processing data volumes, thereby realizing the screening of data processing groups with lower energy consumption, and then realizing effective control of the energy consumption of data processing tasks, thereby reducing the energy consumption in the data processing process.

[0015] A further technical solution is that the data processing requirements are determined according to the data processing tasks, specifically including bad data identification, data conversion and model training.

[0016] A further technical solution is that a method for determining similar processing tasks of the data processing tasks is as follows:

[0017] Based on the data volume of different data types, the deviation amount of the data volume of different data types and different historical processing tasks is determined, and the data similarity coefficient is determined by using the deviation amount of the data volume of different data types;

[0018] Determine the deviation of data volume under different data processing requirements based on data processing requirements of different data types, and determine the data processing similarity coefficient using the deviation of data volume under different data processing requirements;

[0019] The task processing similarity coefficient is determined according to the data processing similarity coefficient and the data similarity coefficient, and the similar processing tasks in the data processing tasks are determined using the task processing similarity coefficient.

[0020] A further technical solution is that the data similarity coefficient is determined based on an average value of the ratio of the deviation amount to the data amount in different data types.

[0021] A further technical solution is that the data processing similarity coefficient is determined based on the number of data processing requirements whose data volume deviation is within a preset deviation range, and specifically the ratio of the number of data processing requirements whose data volume deviation is within the preset deviation range to the number of data processing requirements of the data processing task is determined.

[0022] A further technical solution is that the method for determining the data processing group in the available group is:

[0023] Based on the historical energy consumption data of the data centers in the available group, determine the historical energy consumption data of different data centers in the available group under different processing data volumes;

[0024] According to the historical energy consumption under different processing data volumes, the energy consumption deviation coefficient under different processing data volumes is determined, and the energy consumption coefficient abnormal values ​​of different data centers are determined by the average value of the energy consumption deviation coefficient under different processing data volumes;

[0025] The energy consumption comprehensive abnormal value of the available group is determined by summing the energy consumption coefficient abnormal values ​​of different data centers in the available group, and the data processing group in the available group is determined by using the energy consumption comprehensive abnormal value.

[0026] A further technical solution is that the energy consumption deviation coefficient is determined based on the ratio of the historical energy consumption under the processing data volume to the preset energy consumption under the processing data volume.

[0027] A further technical solution is that the data processing group is an available group with the smallest comprehensive abnormal value of energy consumption.

[0028] Other features and advantages will be described in the following description. The objects and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and drawings.

[0029] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings.

[0031] Figure 1 is a flow chart of a data center cluster scheduling method for energy-saving control;

[0032] Figure 2 is a flow chart of a method for determining similar processing tasks of a data processing task;

[0033] Figure 3 is a flow chart of a method for determining data processing reliability in a data center;

[0034] Figure 4 is a flow chart of a method of determining an available group in a data center group;

[0035] Figure 5 is a flow chart of a method for determining a data processing group among available groups. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0037] During the scheduling and processing of data centers, the present application comprehensively considers the data interaction and energy consumption between different data centers, and performs scheduling and processing of data centers for data processing tasks, which not only reduces the data interaction between different data centers, but also realizes energy-saving control of data centers.

[0038] The task processing similarity coefficient between the data processing task and different historical processing tasks is determined according to the average value of the deviation coefficient of the data volume in different data types, and the historical processing tasks with task processing similarity coefficients greater than 0.6 are regarded as similar processing tasks.

[0039] Based on the data processing delay and energy consumption of similar processing tasks in each data center, if there is no data center with a data processing delay less than a preset processing delay and an energy consumption less than a preset energy consumption, it is determined that there is no data center whose data processing reliability meets the requirements.

[0040] A data center group is constructed using multiple data centers that simultaneously perform historical processing data of similar processing tasks. When similar processing personnel are performing processing, the data center group in which the number of data interactions between different data centers is less than the preset number of interactions is used as an available group.

[0041] The historical energy consumption data of different data centers in the available group are used to determine the average daily energy consumption of different data centers, and the data processing group for task processing of the data processing task is determined based on the minimum value of the sum of the average daily energy consumption of different data centers.

[0042] Specifically, according to one aspect of the present invention, Figure 1 As shown, a first aspect is provided. The present invention provides a data center cluster scheduling method for energy-saving control, which specifically includes:

[0043] S1 determines the data volume of different data types corresponding to the data processing task based on the analysis result of the data processing task, and determines the similar processing tasks and task processing similarity coefficients of the data processing task in combination with the data processing requirements of different data types;

[0044] Furthermore, the data processing requirements are determined according to the data processing tasks, specifically including bad data identification, data conversion and model training.

[0045] Specifically, Figure 2 As shown, the method for determining similar processing tasks of the data processing task is:

[0046] Based on the data volume of different data types, the deviation amount of the data volume of different data types and different historical processing tasks is determined, and the data similarity coefficient is determined by using the deviation amount of the data volume of different data types;

[0047] Determine the deviation of data volume under different data processing requirements based on data processing requirements of different data types, and determine the data processing similarity coefficient using the deviation of data volume under different data processing requirements;

[0048] The task processing similarity coefficient is determined according to the data processing similarity coefficient and the data similarity coefficient, and the similar processing tasks in the data processing tasks are determined using the task processing similarity coefficient.

[0049] Optionally, the data similarity coefficient is determined based on an average value of a ratio of a deviation amount to a data amount in different data types.

[0050] It should be noted that the data processing similarity coefficient is determined based on the number of data processing requirements whose data volume deviation is within a preset deviation range. Specifically, it is determined by the ratio of the number of data processing requirements whose data volume deviation is within the preset deviation range to the number of data processing requirements of the data processing task.

[0051] Furthermore, the similar processing tasks in the data processing tasks are data processing tasks whose task processing similarity coefficient is greater than a preset similarity coefficient.

[0052] In another embodiment, a method for determining similar processing tasks of the data processing task is:

[0053] Acquire the data processing requirements of the historical processing tasks, determine that the data processing requirements of the historical processing tasks do not have data processing requirements that are consistent with those of the data processing tasks, and then determine that the historical processing tasks do not belong to similar processing tasks;

[0054] When there is a data processing requirement consistent with the data processing task:

[0055] The data processing requirements consistent with the data processing task are regarded as consistent processing requirements, and when the number of the consistent processing requirements does not meet the requirement, it is determined that the historical processing task does not belong to the similar processing task;

[0056] When the number of consistent processing requirements is met:

[0057] Based on the data volume of different data types, determine the deviation between the data volume of different data types and different historical processing tasks. When there is no data volume deviation that meets the requirements of the data type:

[0058] It is determined that the historical processing task does not belong to a similar processing task;

[0059] When there is a deviation in the amount of data that meets the required data type:

[0060] Determine a data similarity coefficient by using the deviation of the data amount of different data types, and when the data similarity coefficient does not meet the requirement, determine that the historical processing task does not belong to the similar processing task;

[0061] When the data similarity coefficient meets the requirements:

[0062] Obtaining the amount of processed data under different consistent processing requirements, and when there is no consistent processing requirement that satisfies the processing data amount, determining that the historical processing task does not belong to a similar processing task;

[0063] When there is a consistent processing requirement with the amount of processed data meeting the requirements:

[0064] Determine the data processing similarity coefficient by using the deviation of the data volume under different consistent processing requirements, and when the data processing similarity coefficient does not meet the requirements, determine that the historical processing task does not belong to the similar processing task;

[0065] When the data processing similarity coefficient meets the requirements:

[0066] The task processing similarity coefficient is determined according to the data processing similarity coefficient and the data similarity coefficient, and the similar processing tasks in the data processing tasks are determined using the task processing similarity coefficient.

[0067] S2 determines the data processing delay and energy consumption of different similar processing tasks in each data center, and when it is determined that there is no data center that meets the data processing reliability requirements based on different task processing similarity coefficients, proceeds to the next step;

[0068] Specifically, Figure 3 As shown, the method for determining the data processing reliability of the data center is:

[0069] Based on the data processing delay and energy consumption of different similar processing tasks in the data center, determine the number of historical processing times in which the data processing delay is not greater than the preset processing delay and the energy consumption is not greater than the preset energy consumption, and use it as the reliable processing number;

[0070] Normalizing the task processing similarity coefficients of similar processing tasks corresponding to different historical processing times, and determining similarity weight coefficients corresponding to different historical processing times;

[0071] The data processing reliability of the data center is determined according to the sum of similarity weight coefficients of different reliable processing times.

[0072] Furthermore, the data processing reliability of the data center ranges from 0 to 1, wherein when the data processing reliability of the data center is greater than a preset reliability threshold, it is determined that the data processing reliability of the data center meets the requirements.

[0073] It can be understood that when there is a data center whose data processing reliability meets the requirements, the data center with the highest data processing reliability is used to perform data processing for the data processing task.

[0074] In another embodiment, the method for determining the data processing reliability of the data center is:

[0075] Determine the recommended processing delay of the data center by taking the average value of the product of the data processing delay of different similar processing tasks in the data center and the task processing similarity coefficient;

[0076] Determine the recommended processing energy consumption of the data center by taking the average value of the product of the energy consumption of different similar processing tasks in the data center and the task processing similarity coefficient;

[0077] The data processing reliability of the data center is determined according to the recommended processing energy consumption and the recommended processing delay of the data center.

[0078] In another embodiment, the method for determining the data processing reliability of the data center is:

[0079] Based on the recommended processing energy consumption and the recommended processing delay of the data center, determining a preset processing reliability under the recommended processing energy consumption and a preset processing reliability under the recommended processing delay;

[0080] The data processing reliability of the data center is determined according to an average value of the preset processing reliability under the recommended processing energy consumption and the preset processing reliability under the recommended processing delay.

[0081] Optionally, the method for determining the data processing reliability of the data center is:

[0082] S21 obtains the energy consumption of different similar processing tasks in the data center, and determines the processing abnormality coefficients of different similar processing tasks in combination with the processing delay in the data center, uses the processing abnormality coefficients to determine the abnormal processing tasks among the similar processing tasks, and determines the basic abnormality coefficient according to the proportion of the abnormal processing tasks in the similar processing tasks;

[0083] Optionally, the above step S21 includes the following contents:

[0084] S211 obtains the energy consumption and processing delay of different similar processing tasks in the data center. When the average value of the energy consumption or the average value of the processing delay of different similar processing tasks in the data center does not meet the requirements, it is determined that the data processing reliability of the data center does not meet the requirements. When the average value of the energy consumption or the average value of the processing delay of different similar processing tasks in the data center meets the requirements, it proceeds to step S212;

[0085] S212 determines the number of historical processing times in which the data processing delay is greater than the preset processing delay or the energy consumption is greater than the preset energy consumption based on the data processing delay and energy consumption of different similar processing tasks in the data center, and uses it as the number of abnormal processing times. When the number of abnormal processing times does not meet the requirements, it is determined that the data processing reliability of the data center does not meet the requirements. When the number of abnormal processing times meets the requirements, it proceeds to step S213.

[0086] S213 obtains the energy consumption of different similar processing tasks in the data center, and determines the processing abnormality coefficients of different similar processing tasks in combination with the processing delay in the data center. When the average value of the processing abnormality coefficients of different similar processing tasks does not meet the requirements, it is determined that the data processing reliability of the data center does not meet the requirements. When the average value of the processing abnormality coefficients of different similar processing tasks meets the requirements, it proceeds to step S214;

[0087] S214 uses the processing exception coefficient to determine the exception processing tasks among the similar processing tasks, and determines the basic exception coefficient according to the proportion of the exception processing tasks in the similar processing tasks. When the basic exception coefficient does not meet the requirements, it is determined that the data processing reliability of the data center does not meet the requirements. When the basic exception coefficient meets the requirements, it proceeds to step S22.

[0088] S22, based on the task processing similarity coefficient of the exception processing task, determining the average value of the task processing similarity coefficients of different exception processing tasks, and using the average value as the similarity coefficient average value;

[0089] Optionally, the above step S22 includes the following contents:

[0090] S221 determines the exception weight coefficients of different exception handling tasks according to the task processing similarity coefficients of different exception handling tasks. When the sum of the exception weight coefficients of different exception handling tasks does not meet the requirements, it is determined that the data processing reliability of the data center does not meet the requirements. When the sum of the exception weight coefficients of different exception handling tasks meets the requirements, it proceeds to step S222;

[0091] S222 determines the modified abnormality coefficients of different abnormality processing tasks by multiplying the abnormality weight coefficients of different abnormality processing tasks by the basic abnormality coefficients. When there is an abnormality processing task whose modified abnormality coefficient is greater than the preset coefficient threshold, it is determined that the data processing reliability of the data center does not meet the requirements. When there is no abnormality processing task whose modified abnormality coefficient is greater than the preset coefficient threshold, it proceeds to step S223.

[0092] S223 uses the task processing similarity coefficient of the exception processing task as a basis to determine the average value of the task processing similarity coefficients of different exception processing tasks, and uses it as the similarity coefficient average value, and then proceeds to step S23.

[0093] S23 determines the data processing reliability of the data center based on the difference between a preset value and the product of the basic abnormality coefficient and the similarity coefficient average value.

[0094] S3 determines data center groups corresponding to different historical processing times by using the historical processing conditions of the similar processing tasks in each data center, and determines available groups in the data center group based on the data interaction conditions between data centers in the historical processing times of the similar processing tasks in different data center groups;

[0095] Furthermore, the data center group is a group constructed from data centers that were used to process similar processing tasks in the history.

[0096] Specifically, the data interaction situation between the data centers includes the number of data interactions between different data centers and the amount of interaction data at different numbers of data interactions.

[0097] Specifically, Figure 4 As shown, the method for determining the available groups in the data center group is:

[0098] Using the historical processing times of the data center group for the similar processing tasks as reference processing times;

[0099] Determine the total number of data interactions in different reference processing times according to data interaction conditions between data centers in different reference processing times;

[0100] The average value of the number of interactions of the data center group is determined by the average value of the total number of data interactions in different reference processing times, and the available groups in the data center group are determined by using the average value of the number of interactions.

[0101] Furthermore, the available group is a data center group whose average number of interactions is less than a preset number of interactions.

[0102] In another embodiment, the method for determining the available groups in the data center group is:

[0103] Using the historical processing times of the data center group for the similar processing tasks as reference processing times;

[0104] According to the data interaction conditions between data centers in different reference processing times, the total number of data interactions in different reference processing times is determined, and the reference processing times with the total number of data interactions greater than the preset interaction number threshold are used as the interaction busy processing times;

[0105] An available group in the data center group is determined according to the number of interactive busy processing times.

[0106] Furthermore, the available group in the data center group is a data center group whose interactive busy processing times are less than a preset processing times.

[0107] Optionally, the method for determining the available groups in the data center group is:

[0108] S31 uses the historical processing times of the data center group in the similar processing task as the reference processing times, and determines the total data interaction times in different reference processing times according to the data interaction conditions between the data centers in different reference processing times;

[0109] Optionally, the above step S31 includes the following contents:

[0110] S311 uses the historical processing times of the data center group in the similar processing tasks as the reference processing times, determines the total data interaction times in different reference processing times according to the data interaction situations between the data centers in different reference processing times, and determines the average interaction times of the data center group by the average of the total data interaction times in different reference processing times. When the average interaction times does not meet the requirements, it is determined that the data center group does not belong to the available group. When the average interaction times meets the requirements, it proceeds to step S312.

[0111] S312: When there is no reference processing number whose total number of data interactions is greater than the preset interaction number threshold, the process proceeds to step S313; when there is a reference processing number whose total number of data interactions is greater than the preset interaction number threshold, the process proceeds to step S314;

[0112] S313 obtains the number of data centers with data interaction in different reference processing times. When the number of data centers without data interaction is greater than the reference processing times of the preset center number threshold, the data center group is determined to be an available group. When the number of data centers with data interaction is greater than the reference processing times of the preset center number threshold, the process proceeds to step S32.

[0113] S314: When the reference processing times when the number of data interactions is greater than the preset interaction times threshold value does not meet the requirements, it is determined that the data center group does not belong to the available group; when the reference processing times when the number of data interactions is greater than the preset interaction times threshold value meets the requirements, proceed to step S32.

[0114] S32 determines the interaction processing frequency coefficients of different reference processing times based on the total number of data interactions of different reference processing times and the number of interactions between different data centers;

[0115] Optionally, the above step S32 includes the following contents:

[0116] S321 determines the interaction processing frequency coefficients of different reference processing times based on the total number of data interactions of different reference processing times and the number of interactions between different data centers. When there is a reference processing number with an interaction processing frequency coefficient greater than the preset frequency coefficient, the process proceeds to step S322. When there is no reference processing number with an interaction processing frequency coefficient greater than the preset frequency coefficient, it is determined that the data center group belongs to an available group.

[0117] S322: taking the reference processing times whose interaction processing frequency coefficient is greater than the preset frequency coefficient as the frequent processing times; when the frequent processing times do not meet the requirement, determining that the data center group does not belong to the available group; when the frequent processing times meet the requirement, proceeding to step S323;

[0118] S323 calculates the sum of the interaction processing frequency coefficients of the frequent processing times according to the interaction processing frequency coefficients of different frequent processing times. When the sum of the interaction processing frequency coefficients of the frequent processing times is greater than the preset processing frequency coefficient threshold, it is determined that the data center group does not belong to the available group. When the sum of the interaction processing frequency coefficients of the frequent processing times is not greater than the preset processing frequency coefficient threshold, proceed to step S33.

[0119] S4 determines the historical energy consumption data of different data centers in the available group based on the composition data of the data centers in the available group, and uses the historical energy consumption data to determine the data processing group in the available group, and performs data processing of the data processing task through the data processing group.

[0120] S33 determines the average frequent coefficient of the data center group by average values ​​of the frequent coefficients of the interactive processing with different reference processing times, and determines the available groups in the data center group by using the average frequent coefficient.

[0121] Furthermore, the historical energy consumption data of the data center is determined according to the historical energy consumption under different processing data volumes.

[0122] Specifically, Figure 5 As shown, the method for determining the data processing group in the available group is:

[0123] Based on the historical energy consumption data of the data centers in the available group, determine the historical energy consumption data of different data centers in the available group under different processing data volumes;

[0124] According to the historical energy consumption under different processing data volumes, the energy consumption deviation coefficient under different processing data volumes is determined, and the energy consumption coefficient abnormal values ​​of different data centers are determined by the average value of the energy consumption deviation coefficient under different processing data volumes;

[0125] The energy consumption comprehensive abnormal value of the available group is determined by summing the energy consumption coefficient abnormal values ​​of different data centers in the available group, and the data processing group in the available group is determined by using the energy consumption comprehensive abnormal value.

[0126] Furthermore, the energy consumption deviation coefficient is determined according to a ratio of a historical energy consumption under the processing data volume to a preset energy consumption under the processing data volume.

[0127] It should be noted that the data processing group is the available group with the smallest comprehensive abnormal value of energy consumption.

[0128] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0129] The above is a description of a specific embodiment of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0130] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. A data center cluster scheduling method for energy-saving control, characterized in that: Specifically include: Determine the data volume of different data types corresponding to the data processing task based on the analysis result of the data processing task, and determine the similar processing tasks and task processing similarity coefficients of the data processing task in combination with the data processing requirements of different data types; Determine the data processing delay and energy consumption of different similar processing tasks in each data center, and when it is determined that there is no data center that meets the data processing reliability requirements based on different task processing similarity coefficients, proceed to the next step; Using the historical processing conditions of the similar processing tasks in each data center, data center groups corresponding to different historical processing times are determined, and the available groups in the data center group are determined based on the data interaction conditions between data centers in the historical processing times of the similar processing tasks in different data center groups; Determine historical energy consumption data of different data centers in the available group based on composition data of the data centers in the available group, and use the historical energy consumption data to determine a data processing group in the available group, and perform data processing of data processing tasks through the data processing group; The method for determining the available groups in the data center group is: Using the historical processing times of the data center group for the similar processing tasks as reference processing times; Determine the total number of data interactions in different reference processing times according to data interaction conditions between data centers in different reference processing times; Determine an average value of the number of interactions of the data center group by averaging the total number of data interactions in different reference processing times, and determine an available group in the data center group by using the average number of interactions; The available group is a data center group whose average number of interactions is less than a preset number of interactions.

2. The data center cluster scheduling method for energy-saving control according to claim 1, characterized in that: The data processing requirements are determined according to the data processing tasks, and specifically include bad data identification, data conversion, and model training.

3. The data center cluster scheduling method for energy-saving control according to claim 1, characterized in that: The method for determining similar processing tasks of the data processing task is: Based on the data volume of different data types, the deviation amount of the data volume of different data types and different historical processing tasks is determined, and the data similarity coefficient is determined by using the deviation amount of the data volume of different data types; Determine the deviation of data volume under different data processing requirements based on data processing requirements of different data types, and determine the data processing similarity coefficient using the deviation of data volume under different data processing requirements; The task processing similarity coefficient is determined according to the data processing similarity coefficient and the data similarity coefficient, and the similar processing tasks in the data processing tasks are determined using the task processing similarity coefficient.

4. The data center cluster scheduling method for energy-saving control according to claim 3, characterized in that: The data similarity coefficient is determined according to an average value of the ratio of the deviation amount to the data amount in different data types.

5. The data center cluster scheduling method for energy-saving control according to claim 3, characterized in that: The data processing similarity coefficient is determined based on the number of data processing requirements whose data volume deviation is within a preset deviation range, and specifically determined by the ratio of the number of data processing requirements whose data volume deviation is within the preset deviation range to the number of data processing requirements of the data processing task.

6. The data center cluster scheduling method for energy-saving control according to claim 1, characterized in that: The similar processing tasks in the data processing tasks are data processing tasks whose task processing similarity coefficient is greater than a preset similarity coefficient.

7. The data center cluster scheduling method for energy-saving control according to claim 1, characterized in that: The data center group is a group of data centers that were historically used to process similar processing tasks.

8. The data center cluster scheduling method for energy-saving control according to claim 1, characterized in that: The data interaction situation between data centers includes the number of data interactions between different data centers and the amount of interaction data at different numbers of data interactions.

Images